fix some spelling error with applications/Chat/examples/ (#3692)

* fix spelling error with examples/comminity/

* fix spelling error with example/

applications/Chat/examples/README.md

@@ -24,7 +24,6 @@
     - [LLaMA](#llama)
   - [Add your own models](#add-your-own-models)
     - [Actor model](#actor-model)
-    - [LM model](#lm-model)
     - [Reward model](#reward-model)
     - [Critic model](#critic-model)
 
@@ -150,7 +149,7 @@ torchrun --standalone --nproc_per_node=4 train_prompts.py \
          --strategy colossalai_zero2 \
          --prompt_dataset /path/to/your/prompt_dataset \
          --pretrain_dataset /path/to/your/pretrain_dataset \
-         --rm_pretrain /your/pretrain/rm/defination \
+         --rm_pretrain /your/pretrain/rm/definition \
          --rm_path /your/rm/model/path
 ```
 

applications/Chat/examples/community/peft/easy_dataset.py

@@ -188,7 +188,7 @@ class EasySFTDataset(Dataset):
                 else:
                     raw_input_ids.append(encoded_ids)
 
-        grouped_inpup_ids = []
+        grouped_input_ids = []
         current_input_ids = []
         attention_mask = []
         if tokenizer.pad_token_id is None:
@@ -199,7 +199,7 @@ class EasySFTDataset(Dataset):
                     #pad the current_input_ids to max_length with tokenizer.pad_token_id
                     padded_length = max_length - len(current_input_ids)
                     current_input_ids.extend([tokenizer.pad_token_id] * padded_length)
-                    grouped_inpup_ids.append(torch.tensor(current_input_ids, dtype=torch.long))
+                    grouped_input_ids.append(torch.tensor(current_input_ids, dtype=torch.long))
                     attention_mask.append(
                         torch.tensor([1] * (max_length - padded_length) + [0] * padded_length, dtype=torch.long))
                     current_input_ids = []
@@ -208,7 +208,7 @@ class EasySFTDataset(Dataset):
             if len(current_input_ids) > 0:
                 padded_length = max_length - len(current_input_ids)
                 current_input_ids.extend([tokenizer.pad_token_id] * padded_length)
-                grouped_inpup_ids.append(torch.tensor(current_input_ids, dtype=torch.long))
+                grouped_input_ids.append(torch.tensor(current_input_ids, dtype=torch.long))
                 attention_mask.append(
                     torch.tensor([1] * (max_length - padded_length) + [0] * padded_length, dtype=torch.long))
         else:
@@ -218,8 +218,8 @@ class EasySFTDataset(Dataset):
                 input_ids.extend([tokenizer.pad_token_id] * padded_length)
                 attention_mask.append(
                     torch.tensor([1] * (max_length - padded_length) + [0] * padded_length, dtype=torch.long))
-                grouped_inpup_ids.append(torch.tensor(input_ids, dtype=torch.long))
-        self.input_ids = grouped_inpup_ids
+                grouped_input_ids.append(torch.tensor(input_ids, dtype=torch.long))
+        self.input_ids = grouped_input_ids
         self.labels = copy.deepcopy(self.input_ids)
         self.file_name = data_file
         self.attention_mask = attention_mask

applications/Chat/examples/community/peft/train_peft_prompts.py

@@ -41,7 +41,7 @@ def main(args):
     # configure model
     if args.model == 'bloom':
         # initial_model = BLOOMActor(pretrained=args.pretrain)
-        print('Using peft lora to load Bloom model as inital_model')
+        print('Using peft lora to load Bloom model as initial_model')
         initial_model = BLOOMActor(pretrained=args.pretrain, lora_path=args.sft_lora_path)
         print('Using peft lora to load Bloom model as initial_model (Done)')
     else:

applications/Chat/examples/community/peft/train_peft_sft.py

@@ -86,7 +86,7 @@ def train(args):
 
         if args.strategy == 'colossalai_gemini':
             # this is a hack to deal with the resized embedding
-            # to make sure all parameters are ColoParameter for Colossal-AI Gemini Compatiblity
+            # to make sure all parameters are ColoParameter for Colossal-AI Gemini Compatibility
             for name, param in model.named_parameters():
                 if not isinstance(param, ColoParameter):
                     sub_module_name = '.'.join(name.split('.')[:-1])

applications/Chat/examples/train_sft.py

@@ -84,7 +84,7 @@ def train(args):
 
         if args.strategy == 'colossalai_gemini':
             # this is a hack to deal with the resized embedding
-            # to make sure all parameters are ColoParameter for Colossal-AI Gemini Compatiblity
+            # to make sure all parameters are ColoParameter for Colossal-AI Gemini Compatibility
             for name, param in model.named_parameters():
                 if not isinstance(param, ColoParameter):
                     sub_module_name = '.'.join(name.split('.')[:-1])